Antiviral RISC mainly targets viral mRNA but not genomic RNA of tospovirus

Antiviral RNA silencing/interference (RNAi) of negative-strand (-) RNA plant viruses (NSVs) has been studied less than for single-stranded, positive-sense (+)RNA plant viruses. From the latter, genomic and subgenomic mRNA molecules are targeted by RNAi. However, genomic RNA strands from plant NSVs are generally wrapped tightly within viral nucleocapsid (N) protein to form ribonucleoproteins (RNPs), the core unit for viral replication, transcription and movement. In this study, the targeting of the NSV tospoviral genomic RNA and mRNA molecules by antiviral RNA-induced silencing complexes (RISC) was investigated, in vitro and in planta. RISC fractions isolated from tospovirus-infected N. benthamiana plants specifically cleaved naked, purified tospoviral genomic RNAs in vitro, but not genomic RNAs complexed with viral N protein. In planta RISC complexes, activated by a tobacco rattle virus (TRV) carrying tospovirus NSs or Gn gene fragments, mainly targeted the corresponding viral mRNAs and hardly genomic (viral and viral-complementary strands) RNA assembled into RNPs. In contrast, for the (+)ssRNA cucumber mosaic virus (CMV), RISC complexes, activated by TRV carrying CMV 2a or 2b gene fragments, targeted CMV genomic RNA. Altogether, the results indicated that antiviral RNAi primarily targets tospoviral mRNAs whilst their genomic RNA is well protected in RNPs against RISC-mediated cleavage. Considering the important role of RNPs in the replication cycle of all NSVs, the findings made in this study are likely applicable to all viruses belonging to this group.

sequence identity of the nucleocapsid (N) protein and geographic distribution, species of tospoviruses are grouped into an American-and Euro-Asia clade [20,22]. Tomato spotted wilt virus (TSWV) is the type species of Orthotospovirus and categorized into the American clade [19,20]. Tospoviruses have a tripartite RNA genome that, according to their size, are named large (L), medium (M) and small (S) RNA ( Fig 1A). The L RNA segment is of negative polarity and encodes an RNA-dependent RNA polymerase (RdRp) [23]. The M and S RNAs are ambisense, i.e. they encode two non-overlapping open reading frames (ORFs) on opposite strands. The M RNA encodes a precursor for the glycoproteins in the viral complementary (vc) strand, which is further processed into the mature Gn and Gc glycoproteins. These are required for virus particle maturation and presented as spikes on the surface of the particle envelope [24][25][26]. In the viral (v) strand, the M RNA encodes a nonstructural protein (NSm), which is involved in cell-to-cell and long distance movement [27][28][29][30]. The S RNA similarly codes for a second non-structural protein (NSs) in the v strand, and the N protein in the vc strand [31]. The NSs is an RNA silencing suppressor (RSS) and acts by binding dsRNA in a size-independent manner [32][33][34][35]. In binding vsiRNAs, the NSs protein prevents their systemic movement and suppresses systemic RNA silencing as well [36]. The N protein tightly associates with viral genomic RNA and together with a few copies of the RdRp form the (minimal) infectious RNP units to initiate viral replication and transcription [37][38][39]. Viral mRNAs of TSWV and other bunyaviruses are not assembled into RNPs [40][41][42]. The crystal structure of the TSWV N protein has been resolved [43,44] and show an interaction of monomeric N units into trimers. Recently, a reverse genetics (RG) system for TSWV has been established, the first one for a segmented plant NSV, which opens up new possibilities to study the role of viral proteins in the life cycle of TSWV [45].
Antiviral RNA silencing has been extensively studied for (+)ssRNA plant viruses [5,6]. However, less is known on antiviral RNA silencing of plant NSVs. Unlike (+)ssRNA plant viruses that directly use genomic RNA as template for replication, the RNA genome of plant NSVs can only engage in replication/transcription when complexed with N and RdRp into RNPs [46,47]. For plant NSVs RNPs also present the units that move cell-to-cell and spread viral infection long distance [23]. For this reason, viral genomic RNA from NSVs may be well protected against the host antiviral RNAi machinery. In this study, using a tospovirus as a plant segmented NSV model system, antiviral RISC was isolated from TSWV-infected plants and tested on the ability to access and cleave genomic RNA present in viral RNPs in vitro. Furthermore, RISC targeting of viral mRNA and genomic RNA was also tested in planta. Results show that RISCs mainly target tospoviral mRNAs but not genomic RNA embedded within RNPs.

TSWV activated RISC specifically cleaves TSWV S RNA but not GFP RNA in vitro
Unlike with ss(+)RNA plant viruses, viral genomic RNA of plant NSVs are not naked and always associated with N protein to form ribonucleoprotein (RNP) complexes [19,48]. We therefore hypothesized that the viral RNP structure would protect the viral genomic RNA from cleavage by host antiviral RISC. Since viral mRNAs do not associate with N protein into RNPs, we expected that RISC would likely target the viral mRNA. Here we used tospovirus as plant segmented NSV model system to test this hypothesis. The genomic organization and expression strategy of a tospovirus are illustrated in Fig 1A. As a first step, RISC was isolated from TSWV-infected Nicotiana benthamiana using Hydroxyapatite column chromatography followed with Superdex S-200 column Genomic RNA (gRNA) molecules are tightly associated with N protein (in green). Viral transcripts of the five ORFs are aligned below the gRNA strands. Those for the RdRP, Gn/Gc and N (in grey) result from primary transcription of the gRNA, whereas the additional NSm and NSs transcripts from the ambisense M-and S RNA (in orange), respectively, are produced by secondary transcription, i.e. after gRNA has been copied into a complementary strand that serves as a template for secondary transcription (and replication of progeny gRNA). Small dots at the 5' end of the viral mRNAs present heterogenous, host derived, capped-RNA leader sequence resulting from cap-snatching. Minus sign (+) represents the vc strand of genomic RNA. Minus sign (−) and 3 0 to 5 0 designation represent the v strand of genomic RNA. (B) Diagram of antiviral RISC isolation and in vitro cleavage assay for RISC associated nuclease activity. Leaves were harvested from TSWV-infected or mock-treated N. benthamiana plant (~40 g). The extract of leaves chromatography as described previously [12]. To determine which fractions contain RISCassociated ribonuclease activity, every fraction was incubated with DIG-labeled full-length RNA transcripts of TSWV S (Fig 1B). The results showed that the Superdex S-200 gel filtration fractions 4-9 and fractions 11-16 from TSWV-infected plants exhibit RISC-associated nuclease activity to S RNA (S1A Fig). To rule out that the purified N. benthamiana RISC complexes contain nonspecific nuclease activity to S RNA, RISC was also isolated from mock-inoculated N. benthamiana plants and incubated with DIG-labeled full-length S RNA transcripts of TSWV. As shown in S1B Fig, none of the fractions isolated from mock-inoculated plants were able to cleave the S RNA of TSWV. To test the sequence-specific targeting of the (antiviral) RISC-associated nuclease isolated from tospovirus-infected plants, all purified RISC fractions were also incubated in vitro with (heterologous) DIG-labelled RNA transcripts of GFP. The results showed that fractions containing RISC-nuclease activity against TSWV S RNA did not cleave GFP RNA (S1C Fig). Upon repetition of the entire experiment including all controls (and analysed on the same gel), the active RISC fractions again specifically cleaved TSWV S RNA but not GFP RNA in vitro ( Fig 1C).

Viral RNPs protect genomic RNA against cleavage by antiviral RISC in vitro
The crystal structure of the TSWV N-RNA complex as earlier resolved, and representative for RNPs [43,44], indicated that the N protein associates into trimers (Fig 2A) and in which the viral genomic RNA is deeply embedded in a cleft of the N protein ( Fig 2B). To investigate whether the genomic RNA within RNPs was sufficiently protected against RISC-activity isolated from TSWV-infected plants, DIG-labelled full-length S RNA transcripts were first incubated with TSWV N protein to form N-RNA complexes and next incubated with Superdex S-200 fractions. However, fractions 4-9 and fractions 11-16, which earlier cleaved S RNA in vitro, were unable to cleave the viral genomic S RNA complexed with TSWV N protein ( Fig  2C). Although the intensity of bands reflecting S RNA:N complexes became slightly weaker after the addition of fractions 14-16, the majority of S RNA:N complexes was still there in the presence of anti-TSWV RISC. These data suggested that the genomic S RNA becomes well protected against degradation by host antiviral RISC in vitro after association with N protein.
To rule out the possibility that the N protein itself inhibited RISC activity, the experiment was repeated but this time with a previously identified N R94A/R95A mutant compromised in viral RNA binding [39]. When DIG-labeled full-length S RNA transcripts were mixed with high amounts of N R94A/R95A protein, S RNA:N R94A/R95A complexes were formed. However, anti-TSWV RISC fractions were able to degrade the viral S RNA complexed with N R94A/R95A (S2 Fig).
To further characterize the sequence specific targeting of TSWV by antiviral RISC, genomic M RNA of TSWV /or RNA complexed with N protein was subjected to the in vitro RISC cleavage assay. When fraction 6, containing RISC activity (and hereafter named "RISC fraction") was incubated with DIG-labelled full-length M RNA of TSWV, the band of full-length M RNA was completely digested (Fig 2D). When, prior to the incubation with the RISC fraction, DIGlabelled full-length M RNA was incubated with N protein to form N-RNA complexes, the was loaded onto Hydroxyapatite column followed by Superdex S-200 HR gel column. Sixteen Superdex S-200 fractions were finally collected and every fraction (600 μL) was incubated with in vitro transcribed DIG-labelled RNA for the activity of RISC associated nuclease. (C) DIG-labelled full length S genomic RNA was incubated with isolated chromatography fraction 6 (S1A Fig)

PLOS PATHOGENS
Tospoviral mRNAs, but not the gRNAs, are the main targets of antiviral RISC  Fig 2E). These results further strengthened the first observations, made on the S RNA, that antiviral RISC complexes purified from TSWVinfected N. benthamiana were able to in vitro degrade naked TSWV genomic RNA but not when complexed into viral RNPs.

TRV-induced RISC activity against TSWV mainly targets viral mRNA but not genomic RNA of TSWV in planta
Given that RISC was able to access naked TSWV genomic RNA but not the embedded genomic RNA within RNPs in vitro, next TSWV targeting of RISC in planta was analyzed. Recently, using an established reverse genetics system for TSWV, the N protein (resulting after primary transcription from the S gRNA, Fig 1A) was shown to be essential for viral replication and transcription, while the NSs protein (expressed after replication and secondary transcription of the S gRNA, Fig 1A) was dispensable [47]. To assess RISC targeting of TSWV RNA in vivo, an approach was implemented in which N. benthamiana were first infected with tobacco rattle virus (TRV), carrying gene fragments of N and NSs, respectively, to a priori assemble antiviral RISC in planta, and subsequently challenged with TSWV (S3 Fig). We hypothesized that when RISC activated from TRV-N and TRV-NSs targets the genomic v (or vc) TSWV S RNA molecules, leading to their degradation, it would abrogate viral replication and no TSWV infection would be established in N. benthamiana plants. In case RISC only targets the viral mRNA, the outcome of the experiment would depend on the viral gene targeted. Considering the role of N in replication, RISC targeting of N gene transcripts would inhibit virus replication/infection, while targeting those of NSs, being dispensable, would not affect replication and allow the establishment of a viral infection.
To test this hypothesis, a 300 bp gene fragment of N and NSs, respectively, were cloned into TRV. A similar sized fragment of GUS was inserted into TRV and used as a negative control. N. benthamiana plants were agroinoculated with the TRV constructs and after about 10 days, to allow for assembly of antiviral RISC, the local leaves were challenged with TSWV. Whereas at 9 days post inoculation (dpi) typical TSWV symptoms (Fig 3A, arrow) were observed in systemically-infected leaves from the TRV-NSs pre-treated plants and the TRV-GUS control, no symptoms was discerned in the TRV-N pre-treated plants. Western blot and qRT-PCR analysis confirmed the presence of TSWV in systemically-infected leaves of TRV-GUS and TRV-NSs pre-treated plants but the virus was not detected in the TRV-N pre-treated plants (Fig 3B and 3C). Next, the production levels of genomic S RNA and viral N/NSs mRNA were determined in TRV-GUS, TRV-NSs and TRV-N treated plants. To this end, total RNA was extracted and analyzed by northern blot analysis using NSs/N strand-specific probes to detect NSs mRNA and S vRNA molecules (of similar polarity), or N mRNA and S vcRNA (of similar polarity), respectively. In TRV-NSs pre-infected plants the accumulation of NSs mRNA was significantly lower than in TRV-GUS pre-treated plants (Fig 3D and 3F), whereas the accumulation of genomic S vRNA in TRV-NSs pre-infected plants was only slightly lower than in TRV-GUS pre-infected plants (Fig 3D and 3E). When using the strand specific probe for N, genomic S vcRNA and N mRNA was readily detected in TRV-GUS pre-infected plants ( Fig  3G-3I), but not in TRV-N pre-infected plants (Fig 3G-3I). When the experiments were protection assay. (D) In vitro cleavage assay of naked genomic M RNA by the fraction containing RISC-activity. DIG-labelled full length M genomic RNA was incubated with fraction 6 (S1A Fig) containing the RISC-activity. (E) In vitro cleavage assay of TSWV genomic M RNA complexed with N protein by fraction 6 containing antiviral RISC-activity. DIG-labelled full length M genomic RNA was incubated with TSWV N protein to form N-RNA complexes. The RISC fraction was then added into N-RNA complexes to test the RNA protection activity. The signals on the blot were detected by AP-labeled anti-digoxigenin antibodies and followed with BCIP/NBT staining.
https://doi.org/10.1371/journal.ppat.1009757.g002 In analogy to the approach with TRV-N and TRV-NSs, antiviral RISC was tested on the ambisense encoded genes from the TSWV M RNA, i.e. NSm and GP ( Fig 1A). Whereas the viral glycoproteins processed from GP are dispensable for the infection cycle in plants [28,49], NSm is essential for cell-to-cell and long-distance movement and its silencing would lead to a subliminal infection, i.e. restrict the infection to the initially infected cell [28,29,45]. For each gene a 300 bp gene fragment was cloned into TRV, rendering TRV-NSm and TRV-Gn and next used to pre-infect N. benthamiana plants via agro-infiltration, using TRV-GUS as negative control (S3 Fig). At 10 days post agro-infiltration of TRV-NSm, TRV-Gn and TRV-GUS, the local (lower) leaves were challenged with TSWV. Typical TSWV symptoms were detected in the (upper) systemically-infected leaves of TRV-GUS or TRV-Gn pre-treated plants about 9 days post TSWV inoculation ( Fig 4A). However, no viral symptoms were observed in the systemic leaves of plants pre-treated with TRV-NSm. Upon analyses of systemically-infected leaves using western blot and qRT-PCR, N protein and viral RNA was detected in TRV-GUS and TRV-Gn pre-treated plants but not in TRV-NSm pre-treated plants (Fig 4B and 4C). When the TSWV inoculated leaves were analyzed by northern blots using a strand specific probe for NSm, both genomic M vRNA and NSm mRNA were detected in the pre-treated TRV-GUS plants but not in those that were pre-treated with TRV-NSm (Fig 4D-4F). In the case a strand specific probe for Gn was used, genomic M vcRNA and GP mRNA were detected in the TSWV inoculated leaves of both TRV-Gn and TRV-GUS (control) pre-treated plants, however the accumulation of GP mRNA relative to the genomic RNA molecules in TRV-Gn pre-treated plant was significantly lower (Fig 4G-4I). Similar results were obtained when plants were challenged with TSWV 15 days after the TRV-GUS, TRV-NSm or TRV-Gn treatment of N. benthamiana (S4C and S4D Fig). Altogether, these data suggested that RISC, the core effector of RNA silencing, mainly targets tospoviral mRNA.

Purified RISC of a Euro-Asia type tospovirus targets naked genomic RNA in vitro but not when embedded into viral RNPs
To further substantiate the above observations, experiments were repeated but now using Tomato zonate spot virus (TZSV), a distinct representative of the European-Asian type tospoviruses. In analogy to TSWV, RISC was isolated from TZSV-infected leaves using Hydroxyapatite and Superdex S-200 HR gel column chromatography. Pilot experiments revealed that gel   5A). When prior to the incubation with the RISC fraction, the DIG-labelled full-length S RNA transcripts were first incubated with TZSV N protein to form N-RNA complexes in vitro, the addition of the TZSV RISC fraction failed to degrade the S RNA ( Fig 5B). Similar results were obtained when naked DIG-labelled full-length genomic M RNA transcripts of TZSV or complexed with N protein were incubated in the presence of the anti-TZSV RISC fraction (Fig 5C and 5D). These results indicated that TZSV RNPs, like with TSWV, well protect viral genomic RNA against cleavage by antiviral RISC in vitro.

TRV-induced RISC activity against TZSV mainly targets viral mRNA but not genomic RNA of TZSV in planta
To analyze RISC activity against TZSV in planta, a similar approach was applied as described for TSWV. To this end, a 300 bp gene fragment of TZSV N, NSs, NSm, GP and RdRp was cloned into a TRV vector, rendering TRV-N, TRV-NSs, TRV-NSm, TRV-GP and TRV-RdRp, respectively. Next, these clones, including TRV-GUS as negative control, were used to preinfect N. benthamiana plants by agro-infiltration to induce anti-TZSV gene-specific RISC complexes. At 10 days post agro-infiltration, the local leaves of TRV pre-infected plants were inoculated with TZSV. About 9 days after TZSV inoculation typical TZSV symptom were observed in systemically infected leaves of TRV-GUS, TRV-NSs and TRV-Gn pre-treated plants (Fig 6A, arrows). However, no viral symptoms were discerned in systemic leaves of TRV-N, TRV-NSm and TRV-RdRp pre-treated plants (Fig 6A). Western blot and qRT-PCR analyses confirmed the presence of TZSV in the systemically infected leaves of TRV-GUS, TRV-NSs, TRV-Gn pre-infected plants but not in those of TRV-NSm, TRV-N and TRV-RdRp pre-infected plants (Fig 6B and 6C). To further determine the RNA targets of TRV-induced anti-TZSV RISC in planta, total RNA was extracted from the TZSV inoculated leaves of TRV-GUS, TRV-NSs and TRV-N pre-infected plants and analyzed by northern blot using strand specific probes to the respective viral genes. Similar to the results earlier obtained with TSWV in planta, TRV-GUS plants showed the presence of TZSV genomic S vcRNA and N mRNA, but these were absent from TRV-N pre-treated plants (Fig 6D and 6F).

PLOS PATHOGENS
Tospoviral mRNAs, but not the gRNAs, are the main targets of antiviral RISC TRV-GUS control, however levels of NSs mRNA were significantly reduced in those of TRV-NSs (Fig 6G-6I).

The antiviral RISC targets (sub)genomic (m)RNA of CMV, a representative plant positive-sense RNA virus
We next investigated the targets of antiviral RISC to Cucumber mosaic virus (CMV), a representative ss(+)RNA plant virus. CMV contains a tripartite RNA genome of which RNA2 encodes two proteins, 2a and 2b (S6 Fig). While CMV 2a is expressed from the genomic RNA 2 and is essential for viral replication [50], CMV 2b is expressed from a subgenomic mRNA  [51,52]. In analogy to the strategy applied to TSWV and TZSV, TRV carrying CMV 2a and 2b gene fragments were generated to induce anti-CMV RISC in N. benthamiana. We hypothesized that if anti-2a and anti-2b activated RISC would both target CMV genomic RNA 2, no CMV infection would be established. If, alternatively, anti-2b activated RISC would only target and silence subgenomic mRNA 2b, CMV infection would not be abrogated.
TRV clones were generated containing a 300 bp gene fragment of 2a and a 200 bp fragment of 2b of CMV and, next to TRV-GUS as control, used to pre-infect N. benthamiana plants by agro-infiltration. At 10 days post agro-infiltration, the local leaves of TRV pre-infected plants were inoculated with CMV-Fny. While typical CMV symptoms were observed in the systemically-infected leaves of TRV-GUS pre-infected plants 12 days post CMV inoculation, no viral symptoms were observed in any of the systemic leaves from TRV-2a or TRV-2b pre-treated plants (Fig 7A). Western blot and qRT-PCR analyses confirmed the absence of CMV from the systemic leaves of TRV-2a and TRV-2b pre-treated plants (Fig 7B and 7C). Using northern blot analysis CMV genomic vRNA2 was detected in TRV-GUS pre-infected plants using either 2a or 2b strand specific probes, while the strand specific probe for 2b additionally detected the subgenomic RNA2b (Fig 7D-7H). In plants pre-infected with TRV-2a or TRV-2b, neither genomic vRNA2 nor sub-genomic RNA2b were detected (Fig 7D-7H), indicating that for CMV the antiviral RISC is able to target viral (sub)genomic (m)RNA.

Discussion
In plants RNA silencing is part of the host innate immunity system against viruses in which, during viral infection, RISC becomes programmed to specifically identify and degrade the invasive viral RNAs [5,6]. RISC typically targets single stranded RNA in a sequence-dependent manner, by a siRNA guide strand loaded into the RISC complex. Unlike with (+)ssRNA plant viruses, genomic and antigenomic RNAs of plant NSVs are single-stranded and tightly wrapped by N protein into RNPs complexes [19,20,47]. In this study, antiviral RISC activity was analyzed against plant NSVs, to investigate whether antiviral RISC is able to target NSV genomic RNA complexed into RNPs. To this end, fractions containing (antiviral) RISC associated nuclease activity were isolated from N. benthamiana as previously described for TRV-or   Tombusvirus-infected plants [12][13][14], and biochemically analyzed in vitro on the targeting of viral genomic RNA, either naked or complexed with N protein. In addition, and to substantiate the in vitro findings, in planta studies were performed in which antiviral RISC was induced by a pre-infection with TRV containing (partial) gene sequences from the NSV tospoviruses TSWV or TZSV or the (+)ssRNA CMV. The results showed that whereas for CMV all genomic and subgenomic (m)RNA molecules were targeted by RISC, RISC primarily targets viral mRNA from the NSV tospoviruses in vitro and vivo. Genomic (v and vc) RNA complexed with N in vitro, or within RNPs from a natural infection were hardly or not targeted. Altogether, the findings indicate that antiviral RNAi primarily targets tospoviral mRNAs whilst their genomic RNA molecules are well protected in RNPs against RISC-mediated cleavage. Considering the important role of RNPs in the replication cycle of all NSVs, the observations made in this study are likely applicable to all viruses belonging to this group. Whether, in addition, the formation of RNPs also presents a strategy of the virus to evade from antiviral RNAi remains to be debated.
The sequence specific degradation by antiviral RISC was nicely demonstrated in this study as well, where purified antiviral RISC fractions from tospovirus infected N. benthamiana did not target/degrade GFP RNA, while corresponding fractions isolated from mock-treated N. benthamiana did not cleave tospoviral RNA. Previous studies have reported on two different RISC complexes in Arabidopsis thaliana. One complex containing an AGO1 core and sized >500 kDa, and another complex containing AGO4 and sized 60-200 kDa [53]. During the purification of RISC complexes from TSWV-infected N. benthamiana plants two different peak fractions (fractions 4-9 and fractions 12-16) were also observed to exhibit antiviral RISC-activity. However, and in contrast, only one (fractions 2-7) was isolated from TZSV-infected N. benthamiana. RISC fractions isolated from TRV-infected N. benthamiana plants also distributed in two different peaks [13], while no fractions isolated from wild type tobamovirus infected N. benthamiana plant contained RISC-activity. Earlier studies on tombusviruses showed that the RISC-activity could only be detected in fractions collected from TBSV P19 mutant-infected plants but not from wild type virus infected plants [12]. It is not unlikely that all these differences relate to different antiviral RNAi and viral counter defense strategies, although the differences observed between TSWV and TZSV tospoviruses remain a bit puzzling.
Previous studies have shown that tospovirus-specific siRNAs (vsiRNAs) from all three RNA segments are generated during an infection of N. benthamiana and tomato [54,55].  Considering that the genomic RNA strands are well embedded and protected in RNPs against RISC cleavage, tospovirus siRNAs produced likely result from dsRNA precursor structures that originate from the "naked" viral mRNAs. In light of this, the wide distribution of vsiRNAs all over the genome makes sense, since tospoviral mRNA sequences almost span the entire genome. Consistent with the finding that viral mRNA but not genomic (v and vc) RNA is targeted, siRNAs are rarely detected from the intergenic region of S RNA and M RNA segments [54,55]. Although part of the intergenic region is still reflected in the M-and S-RNA transcribed subgenomic mRNAs, and fold into hairpin structures that function in the termination of viral mRNA transcription, they have been proposed to be associated with NSs and/or elements from the translation machinery to engage in translation, which would also protect them against antiviral RNAi targeting.
Findings from this study are also in agreement with earlier studies on pathogen-derived resistance engineered against TSWV [56]. TRV-induced RISC-activation against NSs or GP, both dispensable for viral replication, did not confer resistance against tospoviruses. Genomic RNA replication was still detected and indicated that these strands are not targeted by RISC. This observation is consistent with earlier studies in which pre-infection of apple latent spherical virus carrying NSs gene fragments also did not protect against three different tospoviruses [57]. On the other hand, targeting of N and NSm prevented TSWV infection and is easily explained from their important roles in the life cycle. RISC targeting of only N mRNA would deplete cells from N protein production and abrogate viral replication, while targeting NSm RNA would prevent production of viral movement protein and block local and systemic spread of the virus. Earlier reports have also shown that transgenic expression of N or NSm gene sequences protects against tospovirus infections [56,58]. While transgenic tobacco expressing NSm are immune to TSWV infection at the plant level, protoplasts isolated from these transformants still support replication of TSWV [58], providing further support for the idea that only the (NSm) transcripts, and not the RNPs, are being targeted by RISC, allowing replication to continue but limited to the infected cell.
Studies from other plant NSVs provide support for the idea that antiviral RISC targeting of tospovirus mRNAs, but not RNPs, widely applies to other/all plant NSVs, Rice stripe virus is a tenuivirus from the Phenuiviridae within the order Bunyavidales. Transgenic rice plants that express (partial) sequences of the C3 gene, coding for the nucleocapsid protein, or the C4 gene, encoding the viral movement protein, protect against RSV infection, similarly as observed with tospoviruses. On the other hand, transgenic plants expressing (partial) sequences of p3, encoding a suppressor of RNA silencing, or p4, encoding a non-structural protein of unknown function, did not provide resistance to RSV infection [59]. Proteins p3 and pC3 are encoded in ambisense by the RNA3 segment of RSV, whereas p4 and pC4 genes are encoded in ambisense by the RNA4 segment of RSV. Being encoded from the same RNA segment, their individual silencing and different outcomes on resistance, provides further support for the idea that viral mRNA, but not the genomic (v and vc) RNA present in RNPs, is the primary target of host antiviral RISC. Defense and counter-defense is an endless arms race between plant host and virus. Whether the embedding of genomic (v and vc) RNAs into viral RNPs should/can be regarded as another weapon from the virus to protect against antiviral RISC remains a matter of debate.
In summary, using the segmented NSV tospovirus as model system, we have shown that RISC mainly targets viral mRNA and not genomic RNA. Fig 8 presents an overview summarizing the action of antiviral RISC against discriminate tospovirus RNA molecules produced during an infection cycle. Studies from other plant NSVs provide support for the idea that this is generic for all NSVs.

Plasmid construction
To generate TRV constructs carrying TSWV gene fragments, a 300 bp gene fragment from the N-terminal region of TSWV N, TSWV NSs, TSWV Gn and TSWV NSm was amplified, respectively, from the cDNA that was reverse transcribed from the purified total RNA of N. benthamiana leaves infected with TSWV lettuce isolate, then inserted into vector pTRV2 (YL156) to generate pTRV2-TSWV N, pTRV2-TSWV NSs, pTRV2-TSWV Gn and pTRV2-TSWV NSm.
For generating the constructs pTRV2-CMV 2a and pTRV2-CMV 2b, a 300 bp gene fragment from the N-terminal region of 2a and a 190 bp fragment from C-terminal region of 2b was amplified from cDNA of CMV Fny, then inserted into pTRV2 (YL156).
Generation of constructs pET28a-TSWV N and the expression and subsequent purification of N protein has been described previously [39]. To generate pET28a-TZSV N, the coding region of the N protein gene was amplified from the cDNA of TZSV, inserted into downstream of the 6x His tag in bacterial expression vector pET28a digested with Nde I and EcoR I.
All primers of used in this paper are listed in S1 Table.

PLOS PATHOGENS
Tospoviral mRNAs, but not the gRNAs, are the main targets of antiviral RISC

Virus source and inoculation
Six-week old N. benthamiana plants were used in all agroinfiltration assays. The TSWV isolate from asparagus lettuce (TSWV-LE), TZSV isolate from tobacco and CMV Fny isolate were used in this study [60]. The TSWV, TZSV and CMV isolates were maintained in N. benthamiana. For long-term storage, infected new N. benthamiana leaves were kept in a -80˚C. For plant inoculation, about 1 g fresh infected leaves of N. benthamiana were ground in 10 mL PB buffer (10 mM sodium phosphate, pH 7.0) and this crude extract was mechanically inoculated onto plant leaves. The virus-inoculated N. benthamiana plants were grown in a growth chamber at 25˚C and 16 h light/8 h dark photoperiod.

Hydroxyapatite column and Superdex S-200 column chromatography
The isolation of RISC fractions from virus infected plants was performed as described previously with slight modification [12][13][14]. Briefly, 40 g TSWV-infected, TZSV-infected or mockinoculated N. benthamiana leaf tissue was homogenized in liquid nitrogen and the powder was resuspended in 40 mL loading buffer (10 mM sodium phosphate, pH 6.8). The homogenate was filtered through two layers of gauze and subsequently centrifuge for 15 min twice at 14,000 ×g at 4˚C. Supernatant (40 mL) was loaded onto a 25×3.6 cm 2 column packed with 50 mL Bio-Gel HT Hydroxyapatite (Bio-Rad, Hercules, CA). Chromatography separations were performed on Ä KTA avant 25 (GE Healthcare). After washing the column with 300 mL loading buffer, the bound proteins were eluted using a linear gradient of sodium phosphate buffer (pH 6.8) from 10-200 mM at a flow rate of 1.0 mL/min. Fractions were analyzed for the presence of active RISC by RNA cleavage assay and stored at -80˚C. Active RISC fractions (0.5~1 mL) isolated from Hydroxyapatite column were further fractionated on Superdex 200 Increase 10/300 GL column (GE Hearlthcare, UK) pre-equilibrated with elution buffer (10 mM sodium phosphate, pH 6.8) at a flow rate of 0.5 mL/min. Every fraction (600 μL) was analyzed for RNA cleavage activity and active RISC fractions were stored at -80˚C until further use.

Preparation of RNA transcripts by in vitro transcription, protein expression and purification
RNA transcripts were generated using the T7 Riboprobe in vitro Transcription kit (Promega, Madison, WI). DNA templates was prepared by directly amplifying the full length TSWV S, TSWV M, TZSV S and TZSV M from full length cDNA clones of TSWV and TZSV [45]. The T7 promoter was added into each forward primer so as to generate PCR products containing a T7 promoter. RNA transcripts were synthesized in a reaction mixture containing 10 mM ATP, CTP, GTP (each), 6.5 mM UTP and 3.5 mM DIG-11-UTP (Roche, Germany), 10X T7 RNA Polymerase Buffer (Takara, Japan), 500 ng PCR amplified products as the template (containing a 5' T7 promoter sequence), 20 units of RNase inhibitor, and 40 units of T7 polymerase and incubated at 37˚C for 3 h. The concentrations of RNA transcripts were determined by a Nano-Drop spectrophotometer (Thermo Scientific). Protein expression and purification was performed as described [39]. Constructs pET28a-TSWV N, pET28a-TSWV N R94A/R95A and pET28a-TZSV N were individually transformed into Escherichia coli strain Rosetta (DE3). To express the recombinant proteins, a 10 mL of overnight culture was transferred to a 1-liter culture and incubated at 37˚C in a shaker incubator. When the optical density of culture at OD 600 reached to 0.6, 0.1 mM isopropyl b-D-thiogalactopyranoside was added and incubated overnight at 20˚C to induce protein expression. The purification of 6xHis-tag recombinant proteins was done using nickelnitrilotriacetic acid resin (Ni-NTA, Qiagen, Hilden, Germany) as described [39].

RNA cleavage assays
RNA cleavage reactions were performed by incubating DIG-11-UTP-labeled RNA transcripts or DIG-labeled RNA complexed with N protein fractions from tospovirus-infected plants as described [12]. Five μg N proteins were mixed with 2 μg of DIG-labeled RNA probe in 10 μL volume and the mixtures were incubated for 10 min on ice to form N-RNA complexes. One μL of N-RNA mixtures was taken to incubate with 8 μL chromatography fractions. The mixtures were incubated for 10 min at room temperature to form N-RNA complexes. Cleavage reactions were incubated at 24˚C for 60 minutes in presence of 15 U/mL RNase A inhibitor (TaKaRa, Japan) and 20 mM MgCl 2 . Control reactions were performed with elution buffer or with chromatography fractions from mock-inoculated plants. DIG-labeled RNA transcripts were resolved by electrophoresis on a 1% agarose gel and transferred to a Hybond-N+ membranes (GE Healthcare, UK) using the Bio-Rad semi-dry transfer unit (Bio-Rad, Hercules, CA). DIG-labelled signals on the blots were detected by AP-conjugated anti-digoxigenin antibodies and visualized by incubation in the presence of BCIP/NBT (Sangon Biotech, Shanghai, China) staining.

Western blot analysis
Protein levels of TSWV, TZSV and CMV in N. benthamiana were determined as described [64,65]. Total protein was extracted from systemically-infected leaf tissues of N. benthamiana at 9-12 days post virus inoculation. About 1 g leaf tissue was homogenized in 2 mL extraction buffer (10 mM sodium phosphate, pH 7.0) containing protease inhibitor cocktail (Sigma, Shanghai). Samples were separated on a 10% SDS-PAGE gel and transferred onto a polyvinylidene fluoride (PVDF) membrane (GE Healthcare, UK). To detect TSWV, TZSV and CMV accumulation, the blots were incubated with specific antibodies to TSWV N, TZSV N and CMV CP, followed with a secondary antibody containing HRP-conjugated goat anti-rabbit or goat anti-mouse. The blots were developed using the ECL Substrate Kit (Thermo Scientific, Hudson, NH, USA). ECL signals were visualized using the ChemiDoc Touch Imaging System (Bio-Rad). Blots were also stained with Ponceau S and used as reference for sample loading and quantification. Rabbit polyclonal antibodies specific for TSWV N and TZSV N were produced in this laboratory. Mouse Monoclonal antibody against CMV CP was kindly provided by Prof. Jianxiang Wu at Zhejiang University. HRP-conjugated goat anti-rabbit or goat anti-mouse IgG were purchased from Thermo Fisher Scientific (Invitrogen, Carlsbad, CA, United States).

Quantitative RT-PCR
The RNA levels of TSWV, TZSV and CMV in N. benthamiana were determined using quantitative RT-PCR (qRT-PCR) [66]. Three pairs of primers for each gene were designed to detect and amplify the encoding regions of TSWV N, TZSV N and CMV CP (S1 Table). Quantitative RT-PCR amplifications were performed on a CFX Connect Real-Time System (Bio-Rad, California, U.S.A.) using a Hieff qPCR SYBR Green Master Mix (Yeasen, Shanghai, China). RNA transcripts of TSWV N, TZSV N or CMV CP were in vitro transcribed using linearized recombinant plasmids of pGEM-TSWV N, pGEM-TZSV N or pGEM-CMV CP and quantified using a NanoDrop spectrophotometer (Thermo Scientific). A series of ten-fold dilutions from 1 to 10 −5 ng μL −1 of the RNA transcripts of TSWV N, TZSV N or CMV CP was prepared. The DNA amount of TSWV N, TZSV N or CMV CP in each dilution sample was represented as copies per μL, which was calculated using the following equation: The logarithm of the gene copy number in each dilution sample (x axis) was plotted against the corresponding crossing point value (y axis) to generate the standard curve for TSWV N, TZSV N and CMV CP. The linear regression equation, along with the coefficient of determination (R 2 ), was used to evaluate the standard curves. After the cycle threshold (Ct) of TSWV, TZSV and CMV in infected N. benthamiana was determined, the copy numbers of TSWV N, TZSV N and CMV CP in infected plants were determined from the standard curves based on the crossing points of the samples. The expression of Elongation factor 1-alpha gene (NbEF1α, Niben101Scf08618g01001) was used as an internal control. Each experiment was performed with three biological replicates.

Northern blot analysis
The genomic vRNAs, vcRNA and viral mRNA of TSWV, TZSV and CMV were detected by northern blot as described [45]. To analyze viral RNA on targeting by antiviral RISC induced via TRV carrying tospovirus or CMV gene fragments, total RNA was extracted at 3~5 dpi from TSWV-, TZSV-or CMV-inoculated leaves of N. benthamiana plants that were preinfected with TRV constructs. Five to ten μg total RNA was resolved on a 1% agarose-formaldehyde gel and transferred to Hybond-N+ membranes (GE Healthcare, UK) [45]. A 300 bp gene fragment from the N-terminal region of TSWV and TZSV N, NSs, Gn and NSm was inserted into a TRV vector. To prepare the template for DIG-labeled probes that do not overlap with the gene fragment of TRV constructs, a 450 bp, 500 bp, 500 bp and 500 bp gene fragment from the C-terminal region of TSWV N, TSWV NSs, TSWV GP and TSWV NSm, respectively, was amplified from the full length cDNA clones of TSWV [45]; a 450 bp and 500 bp gene fragment from the C-terminal region of TZSV N and TZSV NSs, respectively, was amplified from TZSV cDNA; a 600 bp gene fragment from the C-terminal region of 2a without overlapping with 2b, and a 170 bp fragment from the C-terminal region of 2b from cDNA clones of CMV-Fny. Due to the small size of CMV 2b, the 2b probe overlapped with the 2b fragment (190 bp) inserted in pTRV2-2b. The T7 promoter was added into each forward primer to generate PCR products containing a T7 promoter. The purified PCR products were used as DNA template for probes. Strand-specific DIG-labeled probes to detect for TSWV N, TSWV NSs, TSWV NSm, TSWV GP, TZSV N, TZSV NSs, CMV 2a and CMV 2b, respectively, were synthesized using a DIG High Prime RNA labeling kit (Roche, Basel, Switzerland). The blots were developed using a DIG-High Prime Detection Starter Kit II (Roche, Switzerland) according to the manufacturer's instructions. Agrobacterium cultures containing pTRV2-NSs, pTRV2-N, pTRV2-NSm or pTRV2-Gn were equally mixed with Agrobacterium cultures containing pTRV1 and infiltrated into 4-leaf stage N. benthamiana plants using a 1 mL needle-less syringe. Ten days after agro-infiltration, these plants were rub inoculated with crude extracts from TSWV-infected leaves. The phenotype of TSWV challenged plants was monitored at 9 dpi.